Implications of recurrent disturbance for genetic diversity

Ian D. Davies1, Geoffrey J. Cary1, Erin L. Landguth2, David B. Lindenmayer1 & Sam C. Banks1
1
The Fenner School of Environment and Society, The Australian National University, Canberra, ACT, Australia
2
Division of Biological Sciences, University of Montana, Missoula, Montana

Keywords Abstract
Allele surfing, disturbance regimes, range
expansion, recolonization, simulation. Exploring interactions between ecological disturbance, species’ abundances and
community composition provides critical insights for ecological dynamics.
Correspondence While disturbance is also potentially an important driver of landscape genetic
Ian D. Davies, The Fenner School of patterns, the mechanisms by which these patterns may arise by selective and
Environment and Society, The Australian neutral processes are not well-understood. We used simulation to evaluate the
National University, Canberra, ACT 2601,
relative importance of disturbance regime components, and their interaction
Australia.
Tel: 61 (0)2 6125 2647;
with demographic and dispersal processes, on the distribution of genetic diver-
Fax: 61 (0)2 6125 0746; sity across landscapes. We investigated genetic impacts of variation in key com-
E-mail: Ian.Davies@anu.edu.au ponents of disturbance regimes and spatial patterns that are likely to respond
to climate change and land management, including disturbance size, frequency,
Funding Information and severity. The influence of disturbance was mediated by dispersal distance
No funding information provided. and, to a limited extent, by birth rate. Nevertheless, all three disturbance regime
components strongly influenced spatial and temporal patterns of genetic diver-
Received: 15 November 2015; Accepted: 23
November 2015
sity within subpopulations, and were associated with changes in genetic struc-
ture. Furthermore, disturbance-induced changes in temporal population
Ecology and Evolution 2016; 6(4): dynamics and the spatial distribution of populations across the landscape
1181–1196 resulted in disrupted isolation by distance patterns among populations. Our
results show that forecast changes in disturbance regimes have the potential to
doi: 10.1002/ece3.1948 cause major changes to the distribution of genetic diversity within and among
populations. We highlight likely scenarios under which future changes to dis-
turbance size, severity, or frequency will have the strongest impacts on popula-
tion genetic patterns. In addition, our results have implications for the
inference of biological processes from genetic data, because the effects of disper-
sal on genetic patterns were strongly mediated by disturbance regimes.

1975; Krebs et al. 2010), critically influence species rich-

Introduction
The importance of ecological disturbance as a driver of
biodiversity patterns is being increasingly recognized as
disturbance regimes change globally (Turner 2010). Glo-
bal change is influencing the frequency, intensity, season-
ality, and spatial patterns of ecological disturbances such
as wildland fires, floods, and hurricanes (Franklin and
Forman 1987; Webster et al. 2005; Boer et al. 2009; Cary
et al. 2012). By definition, disturbance events cause tem-
poral fluctuations in the distribution and abundance of
species by altering niche opportunities (Shea et al. 2004),
with outcomes strongly depending on the event type, spe-
cies life-history strategies (Noble and Slatyer 1980;
Romiguier et al. 2014) and their phenotypic plasticity
(Anderson et al. 2012). Repeated disturbances through
time, otherwise known as the disturbance regime (Gill
ness and composition within ecological communities
(Connell and Slatyer 1977; Sousa 1984; Foster et al. 1998;
Shea et al. 2004; Miller et al. 2011; Gill et al. 2013). How-
ever, the manner in which disturbance regimes affect
genetic diversity is less well-understood (Banks et al.
2013). The possible influence of disturbance regimes on
the spatial and temporal patterns of genetic diversity has
potential ramifications for the viability of populations,
the adaptability of species (Lowe and Allendorf 2010),
and for inferences made about the underlying mecha-
nisms driving landscape genetic structure.
Disturbance regimes are classically considered as the
history of disturbance at a point (Gill 1975; Krebs et al.
2010). This history is a record of components of the dis-
turbance itself without reference to antecedent or postdis-
turbance conditions (Feller 1996; Keeley 2009; Spies et al.

ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 1181
This is an open access article under the terms of the Creative Commons Attribution License, which permits use,
distribution and reproduction in any medium, provided the original work is properly cited.
Disturbance and Genetic Diversity
2012). While these components may differ between
disturbance types, for fire, measures of intensity (Byram
1959), interdisturbance interval, and the season of occur-
rence have relevance for biological outcomes (Gill 1975).
These measures can also be applied to other disturbance
types such as grazing (Koerner and Collins 2014), floods
(Thodsen et al. 2014), pests (Raffa et al. 2008), and
storms (Lirman 2003). While debate exists as to whether
disturbance size (area) should be included as a compo-
nent of disturbance regimes (Krebs et al. 2010) it could
be assumed that larger disturbances will have demo-
graphic and genetic implications by more readily isolating
populations or by mediating the impacts of colonization
processes on spatial genetic patterns (Hallatschek et al.
2007).
Disturbance can influence genetic patterns through both
selection-driven and neutral processes. For instance, fire
regimes influence the evolution of flammability-enhancing
traits in the shrub Ulex parvoflorus (Moreira et al. 2014).
Disturbance regimes can also act as selective “filters” when
related species share traits that influence their susceptibil-
ity to, or requirement of, disturbance (Helmus et al.
2010). Most case studies of the impacts of disturbance on
neutral genetic diversity, although conducted in the con-
text of an historical disturbance regime, have specifically
focussed on quantifying impacts of single disturbance
events only. A synthesis of these studies reveals that genetic
effects are influenced by disturbance severity (e.g., the
number or proportion of survivors in disturbed popula-
tions) and dispersal (Banks et al. 2011). High severity dis-
turbance events can have long-lasting effects on neutral
genetic diversity in isolated populations (Beheregaray et al.
2003), but little impact when rates of in situ survival or
postdisturbance immigration are high (Spear and Storfer
2010; Su
arez et al. 2012; Banks et al. 2015).
While the genetic impacts of disturbance regimes can
be considered as cumulative impacts of past events, it is
important to recognize that the genetic outcomes of any
one disturbance are contingent on disturbance history at
that location. Thus, a diversity of potential states exist
before any particular disturbance event (Hughes and
Connell 1999); states that represent different distribution
and abundance of individuals, and diversity and spatial
structure of genes. This diversity, within and around the
disturbed area, provides the source of migrants and hence
the demographic and genetic context for recovery. Just as
Connell (1980) notes the importance of the “ghost of
competition past” in understanding the diversity and
coevolution of competing species, we argue that patterns
of genetic diversity in the landscape should be examined
in the context of disturbance history.
A history of theoretical research on the genetic
outcomes of extinction-recolonization dynamics in
1182
I. D. Davies et al.
metapopulations (one of many circumstances that can
arise from disturbance regimes) provides a starting point
for developing expectations about the impacts of distur-
bance regimes (Slatkin 1977; Wade and McCauley 1988;
Whitlock and McCauley 1990; Pannell and Charlesworth
2000). Much of this research stems from the seminal
work of Slatkin (1977) who considered the diversity of
migrant sources and population turnover times (spatially
uncorrelated single-colony extinction and recolonization
events) as critical considerations in determining total
metapopulation diversity (HT), diversity within demes
(HS) and differentiation among demes (FST). Slatkin
(1977) identified the differing effects that the so called
propagule and migrant-pool models have on genetic differ-
entiation using a stepping-stone model, noting these as
two extremes of a continuum that exists in nature. A sep-
arate avenue of more recent research has shown that
selectively neutral processes occurring during population
expansion or colonization can lead to spatial genetic
structure that can persist over long time periods (Hal-
latschek et al. 2007; Excoffier and Ray 2008; Waters et al.
2013). This leads to the expectation that genetic patterns
resulting from historical contingency will shape genetic
outcomes of disturbance events occurring within a long-
term regime.
To understand the potentially complex relationships
that may exist between landscape patterns of genetic
diversity and disturbance, we focus on modelling space,
time, demography, disturbance, and genetics in a simple
and neutral form (Gardner et al. 1987). We developed a
simulation model with a key focus on ecologically rele-
vant disturbance regime components that are expected to
vary in response to climate change and environmental
management (Seidl et al. 2011b; Bradstock et al. 2012).
Using this model, we addressed a series of questions
about the impacts and relative importance of these distur-
bance regime components on genetic diversity:
1 How does the size and frequency of disturbance affect
spatial and temporal patterns of genetic diversity across a
landscape? The frequency and size of wildland fire dis-
turbance, for example, is expected to increase in many
regions (Flannigan et al. 2009) and we expect distur-
bance size (i.e., spatially auto-correlated patterns of dis-
turbance events) to be of critical importance.
2 How does variation in the intensity of disturbance medi-
ate the effects of disturbance regimes on genetic diversity?
We consider disturbance as an agent of mortality, that
is otherwise ecologically and genetically neutral, and
represent effects of intensity (severity) by varying the
degree of in situ survival. Extinction of entire popula-
tions within a disturbance footprint represents one
extreme scenario of disturbance intensity (Slatkin 1977;
ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd.
I. D. Davies et al. Disturbance and Genetic Diversity

Wade and McCauley 1988; Whitlock and McCauley

1990; Pannell and Charlesworth 2000). However,
species have diverse mechanisms to persist in situ fol-
lowing disturbance events (Gill 1975; Whelan et al.
2002), and we expect that the impact of disturbance
events on mortality will play an important role by
influencing how predisturbance spatial patterns of
genetic diversity are retained in postdisturbance land-
scapes, and by mediating the influence of recoloniza-
tion on spatial genetic patterns.
3 How do underlying patterns of dispersal and demography
(birth rate) influence the interaction between disturbance
regimes and genetic patterns? We expect that these
parameters will not only influence the patterns of
recovery from specific disturbance events, but influence
the pre-existing spatial genetic context in which recov-
ery occurs.
The novelty of this study was to use simulation to
understand the relative importance of variation in key
components of disturbance regimes and demographic
processes, in driving landscape patterns of genetic diver-
sity. Insights of this nature address a key gap in under-
standing disturbance-driven ecological genetics (Banks
et al. 2013), and help to formulate hypotheses concerning
possible indirect effects on landscape patterns of genetic
diversity through altered disturbance regimes.

Methods

Conceptual approach
We developed a single-species spatially explicit, individ-
ual-based model (Fig. 1) to simulate five key processes
that can potentially influence genetic diversity: (1) disper-
sal distance; (2) birth rate; (3) disturbance frequency; (4)
disturbance size; and (5) severtiy of disturbance effects
(rates of in situ survival) (Table 1). We used a generalized
linear modeling design (ANOVA) (R Core Team, 2014)
to examine the relative importance of these experimental
factors in explaining variation in between- and within-
population genetic diversity. We limited our analysis to
the relative proportion of variance explained by experi-
mental factors and their interactions, rather than report-
ing statistical significance (Cary et al. 2006; White et al.
2014). We also quantified the spatial pattern of genetic
differentiation among subpopulations (isolation-by-dis-
tance, IBD) resulting from different disturbance scenarios.
Landscape
We used a gridded 102 9 102 cell homogeneous land-
scape with each cell assigned the same carrying capacity
(K), after noting that preliminary simulations on a
Figure 1. Flowchart for the genetics, demography and disturbance
simulation model. Model details are fully described in the text and
Appendix S2. Note that in the case of nonoverlapping generations, as
used in this paper, the initial population comprises new born
individuals only. These become adults during the “Age” process.
204 9 204 cell landscape yielded similar final results
(Appendix S1). To ensure a sufficiently large sample size
for analysis, the 102 9 102 cell landscape was divided
into 1156 sample populations, each comprising a square
of nine grid cells. A sample population, therefore, has a
carrying capacity of 9 9 20 (180 individuals). Preliminary
simulations with larger values of K did not substantially
change the final results but did require simulating a
greater number of generations (Appendix S1). All spatial
processes (dispersal and disturbance) were isotropic, as
heterogeneous cost-surfaces for mate selection and disper-
sal were not invoked. This provides the opportunity to
employ a topologically infinite landscape; a torus topol-
ogy (Haefner et al. 1991), which provides three key bene-
fits: (1) measures of genetic difference between
populations by distance do not suffer a sampling bias; (2)
edge effects can be avoided; and (3) populations simu-

ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 1183
I. D. Davies et al.
Ecological representation
Successional dynamics are a well-documented conse-
quence of disturbance (Bengtsson et al. 2000; Lake et al.
2007; Leavesley et al. 2010; Banks et al. 2011), with spe-
cies commonly showing preference for early or late post-
disturbance stages, or having habitat suitability mediated
by disturbance return intervals. Therefore, disturbance
history can have major effects on amount and connectiv-
ity of suitable habitat, as well as viability of populations
(Amarasekare and Possingham 2001). Introducing simula-
tion approaches that explicitly include habitat dynamics
would further improve insights into dynamics of genetic
diversity under variation in disturbance regimes.
In addition, the basic population processes of many
species may not be static in response to disturbance
regimes. For instance, the reproductive strategies of many
plants are tied to disturbance (Gill 1975), and animal dis-
persal behavior and population connectivity may be vari-
able in response to recent disturbance history (Templeton
et al. 2001; Berry et al. 2005; Pereoglou et al. 2013).
These responses will vary with the degree of phenotypic
plasticity exhibited by the species in question (Anderson
et al. 2012).
Disturbance modelling
Significant scope exists to model specific disturbance
types more mechanistically using simulation methods that
vary with respect to complexity, stochasticity and process
representation (Bates and De Roo 2000; Keane et al.
2004; Seidl et al. 2011a). For instance, a consideration of
dispersal distance of organisms suggests that disturbance
shape, among other factors such as fire patchiness and
minimum fire return interval, is also critically important
given characteristics such as length-to-breadth ratio of
disturbance events will influence the rapidity of recolo-
nization dynamics and hence dynamics in genetic diver-
sity.
Assumptions about genetic processes
There are two further areas where this study might be
extended. Firstly, gene flow patterns are strongly tied to
demographic processes in our model, but there are sce-
narios where gene flow can be decoupled from dispersal
and colonization processes where, for example, animal
dispersal is sex-biased or plants show major differences in
seed and pollen dispersal capabilities. Secondly, selection-
driven genetic responses to disturbance history have been
documented (Moreira et al. 2014). Recent theoretical
models have focussed on evolution under spatio-temporal
ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd.
Disturbance and Genetic Diversity
environmental heterogeneity (Blanquart and Gandon
2011), and empirical studies quantifying selective effects
of disturbance regimes on traits and genes would be valu-
able for extending the present study.
Conclusions
Our findings highlight the background conditions under
which species must evolve strategies to persist in an envi-
ronment of recurrent disturbance. Key to this are survival
and dispersal strategies within the context of the temporal
and spatial scale of patterns of disturbance regimes. Dis-
turbance size, intensity, and frequency present conditions
within which life-history parameters, dispersal strategies,
and density-dependent behaviors may serve to limit or
enhance genetic diversity within in situ residual popula-
tions and at the invasion front, as these are the basic
mechanisms driving patterns of genetic diversity in an
environment subject to recurrent disturbance.
Acknowledgments
The authors would like to thank Jacques Gignoux and
Shayne Flint for the development of the software frame-
work in which this model was developed. The software is
available as Appendix S3 and will be available on GitHub
(https://github.com/) in the near future. In addition we
wish to thank Sam Yeaman and three anonymous review-
ers for valuable and insightful suggestions on ways to
improve this work.
Conflict of Interest
None declared.
References
Amarasekare, P., and H. Possingham. 2001. Patch dynamics
and metapopulation theory: the case of successional species.
J. Theor. Biol. 209:333–344.
Anderson, J. T., D. W. Inouye, A. M. McKinney, R. I.
Colautti, and T. Mitchell-Olds. 2012. Phenotypic plasticity
and adaptive evolution contribute to advancing flowering
phenology in response to climate change. Proc. Biol. Sci.
279:3843–3852
Banks, S. C., M. Dujardin, L. McBurney, D. Blair, M. Barker,
and D. B. Lindenmayer. 2011. Starting points for small
mammal population recovery after wildfire: recolonisation
or residual populations? Oikos 120:26–37.
Banks, S. C., G. J. Cary, A. L. Smith, I. D. Davies, D. A.
Driscoll, A. M. Gill, et al. 2013. How does ecological
disturbance influence genetic diversity? Trends Ecol. Evol.
28:670–679.
1193
I. D. Davies et al.
near the mid-range of mammalian microsatellite mutation
rates (Jarne and Lagoda 1996). Initial allele assignment at
each loci was chosen at random (Gibbs 2001; van Strien
et al. 2015) from the ten alleles available for each locus,
such that simulations began with an expected heterozy-
gosity of 0.9.
Disturbance
Disturbance patterns were described by variation in three
factors: (1) disturbance size, (2) disturbance interval; and
(3) minimum disturbance severity (mortality due to dis-
turbance). Disturbances were arbitrarily square in shape
with three levels of size (6, 12, or 24 cells wide). The
place and time of each disturbance was uniformly ran-
dom. Sufficient events of the specified size were generated
to produce a mean interdisturbance interval (at each cell)
of either 30, 60, or 120 generations for the three treat-
ment levels. No constraint was imposed on the minimum
interval between disturbances. The severity of disturbance
effects can vary randomly between cells from 100% down
to a prescribed minimum (100%, 95%, or 85% of indi-
viduals killed by disturbance). For any single simulation,
disturbance size is held constant. We acknowledge that
disturbances commonly occur with intensities or sizes
that have a log distribution (Cumming 2001). Further-
more, their frequency often has a nonuniform distribu-
tion (Boer et al. 2008). However, we have avoided using
typical distributions of disturbance size and frequencies to
unambiguously examine the relative importance of these
two measures. Nevertheless, we examined the conse-
quences of relaxing this constraint in Appendix S1, but
found no change from the results presented here.
Simulation experiment
The simulations proceeded according to Fig. 1, and the
experimental design encompassed all treatment permuta-
tions of variation in disturbance regime and demography
outlined in Table 1. Overall, we considered five factors,
each with three levels (Table 1). The level of each factor
doubled the value of the previous level, which allowed us
to efficiently span a wide domain and capture the shape
of the response variables over treatment levels. We also
performed 12 replicate simulations without disturbance to
provide a baseline for IBD analysis.
Analysis of variance is necessarily sensitive to the range
over which treatment levels are manipulated. In many
cases, reference to real-world circumstances will provide a
guide to a realistic range. However, in the case of a neu-
tral model, which does not specify any particular species,
location or disturbance type, such choices risk being arbi-
trary. Ecological research has shown that the scale-depen-
ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd.
Disturbance and Genetic Diversity
dence of recovery processes (influenced by the rate of
in situ survival and dispersal/recolonization capacity)
mediates the sensitivity of population persistence to dis-
turbance size and frequency (Romme et al. 1998). We
grounded this experiment by bounding the treatment
range at one extreme by choosing values that allow the
population to become low, but not extinct, over the
course of the 5000 generations. At the other extreme, we
chose treatment levels beyond which effects were negligi-
ble. As a guide, it takes 60 generations for the population
to fully recolonize an area impacted by our largest, high
severity disturbance (24 9 24 cells) using the most
restrictive dispersal and demographic scenario (a birth
rate of 3 and a dispersal distance of 0.25). Total abun-
dance over the landscape for this scenario is approxi-
mately 25% of K. Conversely, the same process took only
12 generations under our scenario with highest dispersal
and birth rate (a birth rate of 12 and a dispersal distance
of 1, resulting in a 68% occupancy rate). Viable popula-
tions might also occur if disturbance size is traded against
frequency, and this may be important when considering
the generality of our results. However, increasing birth
rate beyond 12 will have no effect because: (1) the diver-
sity of source populations does not change in a stepping-
stone model (as used here); and (2) birth rates beyond
those used here, even when source populations are very
small, will result in a population at carrying capacity at
the first generation. Therefore, when considering scale
and the application of this general study to particular
cases, it is the relativity between carrying capacity and
disturbance size that should be kept in mind. To confirm
that the disturbance parameters are within a long-term
equilibrium we examined our parameter space using two
metrics (T and S) used to define degrees of landscape
equilibrium (Turner et al. 1993). T is the mean distur-
bance interval divided by the recovery time and S is the
disturbance area divided by the landscape size. The
range of T and S for the above two extreme cases define
a rectangle which lies on the border of “equilibrium or
steady state” and “stable, low variance” landscapes as
defined by Turner et al. (1993). Choosing lower fre-
quency and larger disturbance sizes would be a viable
alternative, (type D in Turner et al. (1993)), however
either proportionally longer simulations would be
required (or more replicates) to limit variance in model
outputs to an acceptable level.
While there are no relevant analytical disturbance mod-
els for comparison, the basic demographic model used
here, when simulated under panmixia, results in patterns
of genetic drift expected under Wright’s formulation of
HT ¼ ð1  ð1=2NÞÞt H0 . Thus, our underlying demo-
graphic-genetic model “delivers” the expectations of pop-
ulation genetic theory, and the dispersal scenarios that we
1185
Disturbance and Genetic Diversity
simulate are sufficiently simple and general that they can
be representative of, or relevant to, many real scenarios.
Statistical analysis of simulation results
Preliminary simulations of 10,000 generations
(Appendix S1, Fig. 6), indicated that HT, H S , and FST
were relatively stable after 4000 generations. Therefore,
each simulation was preceded by a 4000 generation burn-
in with a further 1000 generations executed to provide
data for analysis. We ran 12 replicates for each simulation
resulting in 2916 runs. Replicates differ only in their
random number seeds which affect: (1) initial allele
assignment; (2) Mendelian genetics and mutation; (3) dis-
turbance time and location; (4) distance and direction of
individual dispersal events; and (5) random post-distur-
bance survival. We used only 12 replicates per scenario
because the large landscapes (up to approximately
2 9 105 individuals per generation) resulted in little vari-
ation among replicates within treatments. Indeed, the
coefficient of variation in FST between replicates (averaged
over the last 1000 generations) was, at most, 18% for
simulations with large, frequent and severe disturbances
with low dispersal distance and low growth rates (i.e., sce-
narios with the greatest stochastic variability).
For a given population, heterozygosity was averaged
over the number of loci l for frequency p of allele i of k.
! Effects of disturbance regimes on spatial
1X l Xk
H¼ 1 2
pi (3)
l j¼1 i¼1
The average heterozygosity of populations (H S ) was the
sum of Hi, weighted by the population size (Ni):
Ps
i¼1 Hi Ni
HS ¼ P s (4)
i¼1 Ni
The measure of genetic difference between populations,
where HT was determined using eq. (3) for the entire
population, was:
HT  H S
FST ¼ (5)
HT
In addition, 50 sample populations with N > 0.1 K (18
individuals in nine cells) were randomly selected for pair-
wise FST analysis at the end of each simulation, to quan-
tify spatial genetic structure arising from the experiments.
Pairwise FST for populations A and B is given by (Nei
1978):
HT  ðHA NA þ HB NB Þ=ðNA þ NB Þ
FST ðABÞ ¼ (6)
HT
We quantified the regression of FST/(1–FST) and log-
transformed geographic distance between populations, as
1186
I. D. Davies et al.
suggested by Rousset (1997), and used distograms and
Mantel correlograms to explore the effects of disturbance
regimes on spatial patterning of the magnitude of FST and
the FST by distance correlation, respectively, using the
methods described in Diniz-Filho et al. (2013).
Results
Effects of disturbance regimes on overall
genetic diversity and differentiation
HT was relatively unchanged between experimental treat-
ments compared to ðH S Þ or FST (Fig. 2A). Variation in
HT was driven primarily by the disturbance variables that
directly affect overall population size, especially distur-
bance size, including its interaction with birth rate and
dispersal distance (Fig 2B). Variables that affect popula-
tion size also contribute most to temporal variation in
FST (CVFST), predominantly disturbance size (56% of
variance explained) (Fig. 2C). Because of the relative sta-
bility of HT, there was, by definition (eq. 5), a close
inverse relationship between FST and ðH S Þ (Fig. 2A). For
this reason, and because the major effects of disturbance
were apparent in the spatial patterning of genetic diversity
among subpopulations, we focused predominantly on the
detail of FST in the results (Fig. 2D).
genetic structure (FST)
Variation in disturbance size (DS) explained around 9% of
variation in simulated FST, with only mean dispersal dis-
tance (MD) explaining greater variation (37%), while dis-
turbance interval (DI) and severity explained 7% and 6%
respectively (Fig. 2D). There were important interactions
between the three disturbance variables, (size, frequency,
and severity), and the mean dispersal distance (Fig. 2D).
The effects of disturbance size and frequency were strong
when dispersal distance was restricted to a mean of 0.25
cells, with both causing increases in genetic structure across
the landscape, but the effects of DS and rates of in situ sur-
vial (SV) were negligible when MD was high (mean dis-
tance 1 cell) (Fig. 3A–C).
The least important parameter in our simulations was
birth rate (BR) (Fig. 2D). While birth rate increases the
rate at which the population can regenerate in disturbed
areas, similar to the colonizing effect afforded by high
dispersal distance, our results suggest that, within the
treatment range, birth rate has a far lesser effect on FST in
recovering populations, and therefore on genetic structure
in the presence of disturbance.
Increasing in situ survival rates after disturbance (SV)
markedly reduced any effect of disturbance on genetic
ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd.
I. D. Davies et al. Disturbance and Genetic Diversity

(A) Kernel density estimates (B) AV. HT

200
20
Av. FST DS 0.088
Av. HS DI 0.057
Av. HT SV 0.052
BR 0.026

density Av. HS and Av. FST

MD 0.025

150
15
DS:SV 0.072

Source of variation
DI:DS 0.071

Density Av. HT
DI:SV 0.04
DS:BR 0.037
DS:MD 0.03

100
10
MD:SV 0.029
DI:BR 0.028
DI:MD 0.021
DI:DS:SV 0.058
DS:MD:SV 0.042

50
5
DI:DS:BR 0.041
DI:DS:MD 0.026
DI:MD:SV 0.024
DI:DS:MD:SV 0.035
DI:DS:SV:BR 0.022
0

0
0 0.2 0.4 0.6 0.8 0.0 0.1 0.2 0.3 0.4 0.5
N = 2916 Bandwidth = 0.009594 Relative var. explained (sum 0.98)

(C) CV FST (D) Av. FST

MD 0.366
0.558
DS DS 0.094

DI 0.07

0.088 SV 0.056

Source of variation
Source of variation

SV
BR 0.034
Figure 2. Comparison of (A) kernel density
DS:MD 0.042
estimates of HT, FST and HS from all scenarios 0.032
DI DS:SV 0.039
together with analysis of the relative variation
in three response variables explained by MD:SV 0.036

demographic and disturbance factors. 0.022 DI:MD 0.033

BR
Response variables for ANOVA analysis (R Core 0.03
DI:DS
Team 2014) are relative variance explained for
MD:BR 0.024
(B) Av. HTT, (C) temporal coefficient of 0.084
DS:SV DS:MD:SV 0.021
variation of FST (over 1000 generations) and
(D) Av FST. For clarity, only factors and factor
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.1 0.2 0.3 0.4 0.5
interaction explaining more than 2% of the
variance are shown. Relative var. explained (sum 0.883) Relative var. explained (sum 0.992)

structure, with disturbance having little effect on genetic increased when disturbances were large, frequent and
structure under low severity (SV = 0.85) (Fig. 3E,F). severe (Fig. 4C). However, we found Mantel correlograms
Increasing disturbance interval (DI) decreased the effect more informative of IBD patterns that arose in the
of disturbance size in a linear fashion (not shown). context of disturbance regimes than simply recording
Mantel’s r (Mantel 1967) (Fig 4D). In the absence of dis-
turbance, the scale of positive autocorrelation was stable
Genetic isolation-by-distance
at a distance of approximately 30 cells (the patch size)
In the absence of disturbance in our model system, a across all treatments over the course of the simulations
quasi-linear relationship existed between genetic and geo- (Fig. 4D). However, for scenarios representing highly dis-
graphic distance when the data are transformed as FST/ turbed landscapes, while variability in Mantel’s r by dis-
(1–FST) by ln-distance (Fig. 4A). The slope of this rela- tance class was greatly increased, there was nevertheless a
tionship was sensitive to dispersal distance and also dis- general trend of increasing correlation by distance class
played a levelling off at a distance of log 35 (3.5) cells (both positive and negative) (Fig 4D).
(Fig. 4A). However, this relationship was disrupted in
simulations with high levels of disturbance and low dis-
Temporal variability of FST
persal distance, a situation where populations remain in
fragmented patches for most of the simulation (Fig. 4B). The temporal coefficient of variation in FST (CVFST),
The spread of FST across distance classes was greatly measured over the last 1000 generations of the simula-

ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 1187
Disturbance and Genetic Diversity I. D. Davies et al.

(A) MD:DS (B) MD:DI (C) MD:SV

0.30

0.30

0.30
DS = 6 DI = 120 SV = 0.85
DS = 12 DI = 60 SV = 0.95
DS = 24 DI = 30 SV = 1.0
0.20

0.20

0.20
Av. FST

Av. FST

Av. FST
0.10

0.10

0.10
0.00

0.00

0.00
0.25 0.5 1 0.25 0.5 1 0.25 0.5 1
Dispersal distance (cells) Dispersal distance (cells) Dispersal distance (cells)

(D) BR:DS (E) SV:DS (F) SV:DI

0.30

0.30

0.30
DS = 6 DS = 6 DI = 120
DS = 12 DS = 12 DI = 60
DS = 24 DS = 24 DI = 30
0.20

0.20

0.20
Av. FST

Av. FST

Av. FST
0.10

0.10

0.10
0.00

0.00

0.00
3 6 12 0.85 0.95 1 0.85 0.95 1
Birth rate Severity Severity

Figure 3. Two-way interactions of disturbance treatments and mean dispersal distance (A, B, and C) of variation in FST (averaged over the last 1000
generations). Chart (D) shows the interaction of birth rate and disturbance size. Charts (E) and (F) show the interactions of disturbance severity with
disturbance interval and size. These charts show values averaged over all parameters and replicates, except where otherwise indicated.

tion, was strongly affected by disturbance size, which
explained around 56% of the variance in this response
variable (Fig 2C). However, while higher birth rates
decreased temporal variation of FST, especially for the
highest rate of in situ survival (Fig. 5), variance explained
in CVFST by birth rate was still relatively minor compared
to other parameters (Fig. 2C).
General observations
Informally, for regimes incorporating large, frequent, and
severe disturbances, our model produced a clumped pat-
terns of high and low heterozygosity meditated by disper-
sal distance (Fig. 6). These patterns arose through a
combination of two related pattern-generating processes:
colonization and expansion, whose impacts are mediated
by disturbance severity. As used here, colonization was
defined as spread from a surrounding population into
vacant sites, while range expansion was the opposite;
spreading out from isolated refugia into an empty land-
scape (Fig. 6), This led to patterns of genetic homogene-
ity (low heterozygosity) with zones of higher
heterozygosity where colonization/expansion fronts meet.
Once established, these patterns were relatively stable over
time, and while the size of patches of genetic homogene-
ity were similar over replicates, each replicate was never-
theless unique in detail. The spatial covariance of
variation in HS showed an exponential increase with dis-
turbance size. Using the most restrictive dispersal and
demographic scenarios (a dispersal distance of 0.25 and a
birth rate of 3), and the most severe and frequent distur-
bance (severity of 1 and mean interdisturbance frequency
of 30 years), the spatial coefficient of variation in HS
increased from 2% where disturbance size was zero, to
54% when disturbance size was 24 cells in width. At the
same time, mean HS decreased by 84% from 0.82 to 0.13.

1188 ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd.
I. D. Davies et al. Disturbance and Genetic Diversity

(A) No disturbance (B) Highly disturbed

2.0
MD = 1.0 MD = 1.0
MD = 0.5 MD = 0.5

0.08

1.5
MD = 0.25 MD = 0.25
Figure 4. The relationship between

FST (1−FST)

FST (1−FST)
transformed pairwise FST and log transformed

1.0
distance for (A) undisturbed and (B) highly

0.04
disturbed landscapes. For clarity, the scatter

0.5
data for each replicate simulation is summarized
as a Lowess regression (R Core Team). Chart (C)

0.00

0.0
shows pairwise untransformed FST by distance
class (Diniz-Filho et al. 2013). In chart (D),
1.0 1.5 2.0 2.5 3.0 3.5 4.0 1.0 1.5 2.0 2.5 3.0 3.5 4.0
significant values (P < 0.05) of correlation
ln−distance ln−distance
between genetic and spatial distance matrices
are shown with a “+”, nonsignificant values are
(C) Distogram (D) Correlogram
shown with open circles. The intersection of the
dotted lines indicates the approximate patch No disturbance + No disturbance

0.4
+ +
0.8
Highly disturbed + Highly disturbed
size. These pairwise comparisons use 50, +
+ + + +
+ + + ++ +++
+ +
+ +++
randomly selected sites at the end of each of + ++ ++

0.2
++ + + + +++ + +++++
+ ++
0.6

+ ++ ++ + +++ +
+++ ++ ++

Mantel r
the 12 replicate simulations. Consequently, ++ + ++ +++ +
+ +
+ ++ ++ + +
++ + + + ++
+ ++

0.0
there will be a greater variance between the 12 + + ++ +
+ +
+ + + +
FST

+ ++
0.4

+ ++ ++ ++ + + + + +++ +
++ + +
+ ++
+
replicates shown here, than between the meta- + +++ ++ + + ++ + + ++ ++ + +
++
+ +
++
+ ++
++
+ +
+
+++ +++++ +
+ + + +++ +++ ++++ + +
population FST averaged over 1000 generations + + ++ + + +++ +++ ++
++ +
0.2

++ +++ +
+

—0.4
used for analysis in Fig. 2. All examples use a + + + + +
++ +++ + +
++
+ ++
birth rate of 3 and a dispersal distance of 0.25 ++
+
+ + ++ ++ ++ ++ ++ ++ ++
+ ++ ++
0.0

unless otherwise indicated. For highly disturbed

10 20 30 40 50 60 10 20 30 40 50 60
landscapes, the parameters are DI = 30 years;
DS = 24 cells; SV = 1. Distance class (cells) Distance class (cells)

DS:BR:SV = 1.0 DS:BR:SV = 0.95 DS:BR:SV = 0.85

10

10

10
BR = 12 BR = 12 BR = 12
BR = 6 BR = 6 BR = 6
BR = 3 BR = 3 BR = 3
8

8
6

6
CV FST

CV FST

CV FST
4

4
2

2
0

6 12 24 6 12 24 6 12 24
Disturbance width (cells) Disturbance width (cells) Disturbance width (cells)

Figure 5. Temporal effects on FST of disturbance size, severity and birth rate. Charts show values averaged over all parameters and replicates,
except where otherwise indicated.

et al. 2011). Here, we showed that disturbance regimes

Discussion
can have large impacts on spatial and temporal patterns
of genetic structure. Of all disturbance variables, distur-
Disturbance size and severity influence
bance size and frequency had the greatest genetic effect in
spatial genetic patterns
our model system. However, the amount of in situ
Disturbance regimes are critical attributes of ecosystems survival, (“minimum severity” in our model), strongly
and can have large effects on community composition mediates the overall genetic effects of disturbance
(Lake 2000; Berumen and Pratchett 2006; Lindenmayer (Fig. 4E,F). While the disturbance parameters tested here

ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 1189
Implications of recurrent disturbance for genetic diversity
Ian D. Davies1, Geoffrey J. Cary1, Erin L. Landguth2, David B. Lindenmayer1 & Sam C. Banks1
1
The Fenner School of Environment and Society, The Australian National University, Canberra, ACT, Australia
2
Division of Biological Sciences, University of Montana, Missoula, Montana

Keywords Abstract
Allele surfing, disturbance regimes, range
expansion, recolonization, simulation. Exploring interactions between ecological disturbance, species’ abundances and
community composition provides critical insights for ecological dynamics.
Correspondence While disturbance is also potentially an important driver of landscape genetic
Ian D. Davies, The Fenner School of patterns, the mechanisms by which these patterns may arise by selective and
Environment and Society, The Australian neutral processes are not well-understood. We used simulation to evaluate the
National University, Canberra, ACT 2601,
relative importance of disturbance regime components, and their interaction
Australia.
Tel: 61 (0)2 6125 2647;
with demographic and dispersal processes, on the distribution of genetic diver-
Fax: 61 (0)2 6125 0746; sity across landscapes. We investigated genetic impacts of variation in key com-
E-mail: Ian.Davies@anu.edu.au ponents of disturbance regimes and spatial patterns that are likely to respond
to climate change and land management, including disturbance size, frequency,
Funding Information and severity. The influence of disturbance was mediated by dispersal distance
No funding information provided. and, to a limited extent, by birth rate. Nevertheless, all three disturbance regime
components strongly influenced spatial and temporal patterns of genetic diver-
Received: 15 November 2015; Accepted: 23
November 2015
sity within subpopulations, and were associated with changes in genetic struc-
ture. Furthermore, disturbance-induced changes in temporal population
Ecology and Evolution 2016; 6(4): dynamics and the spatial distribution of populations across the landscape
1181–1196 resulted in disrupted isolation by distance patterns among populations. Our
results show that forecast changes in disturbance regimes have the potential to
doi: 10.1002/ece3.1948 cause major changes to the distribution of genetic diversity within and among
populations. We highlight likely scenarios under which future changes to dis-
turbance size, severity, or frequency will have the strongest impacts on popula-
tion genetic patterns. In addition, our results have implications for the
inference of biological processes from genetic data, because the effects of disper-
sal on genetic patterns were strongly mediated by disturbance regimes.

1975; Krebs et al. 2010), critically influence species rich-

Introduction
The importance of ecological disturbance as a driver of
biodiversity patterns is being increasingly recognized as
disturbance regimes change globally (Turner 2010). Glo-
bal change is influencing the frequency, intensity, season-
ality, and spatial patterns of ecological disturbances such
as wildland fires, floods, and hurricanes (Franklin and
Forman 1987; Webster et al. 2005; Boer et al. 2009; Cary
et al. 2012). By definition, disturbance events cause tem-
poral fluctuations in the distribution and abundance of
species by altering niche opportunities (Shea et al. 2004),
with outcomes strongly depending on the event type, spe-
cies life-history strategies (Noble and Slatyer 1980;
Romiguier et al. 2014) and their phenotypic plasticity
(Anderson et al. 2012). Repeated disturbances through
time, otherwise known as the disturbance regime (Gill
ness and composition within ecological communities
(Connell and Slatyer 1977; Sousa 1984; Foster et al. 1998;
Shea et al. 2004; Miller et al. 2011; Gill et al. 2013). How-
ever, the manner in which disturbance regimes affect
genetic diversity is less well-understood (Banks et al.
2013). The possible influence of disturbance regimes on
the spatial and temporal patterns of genetic diversity has
potential ramifications for the viability of populations,
the adaptability of species (Lowe and Allendorf 2010),
and for inferences made about the underlying mecha-
nisms driving landscape genetic structure.
Disturbance regimes are classically considered as the
history of disturbance at a point (Gill 1975; Krebs et al.
2010). This history is a record of components of the dis-
turbance itself without reference to antecedent or postdis-
turbance conditions (Feller 1996; Keeley 2009; Spies et al.

ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 1181
This is an open access article under the terms of the Creative Commons Attribution License, which permits use,
distribution and reproduction in any medium, provided the original work is properly cited.
I. D. Davies et al.
provides insights about when these theoretical principles
are likely to apply under spatial and temporal variation in
ecological disturbance patterns. For instance, population
turnover rates are expected to increase FST under the
propagule pool model (Pannell and Charlesworth 2000).
This circumstance does not arise where disturbances are
small or dispersal distance is large (relative to disturbance
size), or in situ survivors are common (Gill 1975; Banks
et al. 2011). However, recolonization is a process that oper-
ates in space, and our model suggests that the propagule
pool model will predominate under regimes of severe dis-
turbance when the disturbance footprint is large relative to
dispersal distance. The scale of disturbance size and disper-
sal distance in our study ensures that this occurs even when
the dispersal model is replaced with one that draws from a
larger number of source populations (Appendix S1).
Allele surfing and the role of population
growth rate in spatial genetic patterns
We identified a likely role during disturbance regimes for
the “allele surfing” process documented during range
expansions (Hallatschek et al. 2007; Excoffier and Ray
2008). This was most apparent under scenarios of large
disturbance size and high severity, where recolonization
processes generated patches of genetic homogeneity, with
high heterozygosity occurring in “mixing zones” between
colonization fronts. Such patterns are not seen in undis-
turbed populations, but feature during biological invasions
or extinction-recolonization scenarios driven by distur-
bance regimes, long-term climatic fluctuation and recolo-
nization from refugia (Petit et al. 2003). In our model, this
pattern is dynamic on a generational time scale, with refu-
gia (populations surviving disturbance in situ) forming
and re-joining at a rate dependent on dispersal distance,
disturbance size and the degree of in situ survival. The per-
sistence of these patterns underlies the importance of the
system state before each disturbance. Disturbances that
span a region where two populations of low heterozygosity
meet (a mixing zone) tend to recolonize with the same
pattern of heterozygosity as existed before the disturbance,
leaving relatively little indication of the footprint of the last
disturbance. Having speculated that these patterns of
heterozygosity, once established, appear relatively stable,
future simulation studies may be able to quantify the sta-
bility of these patterns by some measure of their rate of
change. Perhaps it is the case that, after establishing this
“stable” state, the subsequent history of disturbance has, in
fact, far less an effect on landscape genetic patterns than
might be thought.
Our model system employed a neutral, density-inde-
pendent approach to rates of population growth. How-
ever, other research indicates that different outcomes
ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd.
Disturbance and Genetic Diversity
could arise if this assumption were to be relaxed. The
observed allele surfing patterns might not arise in cases
where density-dependent reproduction delays expansion
from founding populations (Waters et al. 2013), or where
gene flow is not tied to demographic expansion (pollen
dispersal). Conversely, even greater rates of genetic drift
during recolonization after severe disturbance will likely
occur if a density-dependent growth function were to be
used that produced higher growth rates at low population
densities (Waters et al. 2013).
The occurrence of allele surfing processes under
regimes of large, severe disturbances may have implica-
tions for the inference of geographical variation selection
from genome-wide association studies. For instance, the
occurrence of disturbance is commonly associated with
spatial environmental variation (Swetnam and Betancourt
2010), but can generate strong spatial heterogeneity in
genetic diversity through selectively-neutral processes such
as allele surfing (Excoffier and Ray 2008).
Our results provide insights into the conditions under
which climate-driven or landscape management-driven
shifts in disturbance patterns will have important conse-
quences for genetic diversity, as influenced by selectively
neutral demographic processes. Changes to disturbance size
or frequency will have strong implications for genetic pat-
terns when severity is high (i.e., there are few survivors)
and dispersal capacity is restricted relative to the scale of
disturbance. This finding has parallels with ecological
research on the scale-dependence of postdisturbance recov-
ery processes, where the spatial dispersion of survivors and
the scaling of colonization ability relative to disturbance
area mediate the sensitivity of population distribution to
disturbance size and return interval (Romme et al. 1998).
Conversely, genetic patterns in species with long-range dis-
persal capability (relative to disturbance size), may be
insensitive to variation in disturbance size or severity. For
many disturbance types, disturbance size and frequency
may be positively correlated (such as wildland fire), and
under such circumstances shifts in disturbance regimes
from global change phenomena like climate change may
have an enhanced role in determining spatial patterns of
genetic diversity across landscapes. For example, King et al.
(2013) have shown that climate change could increase area
burned in mesic forested landscapes while at the same time
reduce area burned in more arid grass dominated regions.
Dispersal patterns mediate disturbance
effects
Dispersal distance had a far stronger effect on H S and FST
than any of the disturbance parameters (Fig. 2D). Fur-
thermore, dispersal distance strongly mediated the
impacts of disturbance regimes on spatial genetic patterns
1191
Disturbance and Genetic Diversity
because, for high dispersal rates, postdisturbance demo-
graphic recovery and genetic diversity were driven
strongly by high immigration, irrespective of the presence
of in situ survivors.
Under more realistic dispersal models (e.g., the nega-
tive exponential model in Appendix S1), higher dispersal
scenarios imply that founders are drawn from larger, and
therefore a more heterogeneous set of source popula-
tions, corresponding to the distinction between the
propagule pool and migrant pool models (Pannell and
Charlesworth 2000). However, when using a stepping-
stone model, the set of source populations can only vary
between one and eight neighbors. Dispersal distance,
therefore, directly determines the size of founding popu-
lations and it is the genetic drift that arises from this
variation that drives the model (Hallatschek et al. 2007;
Excoffier and Ray 2008). However, because we limit the
population to K only after dispersal takes place (Fig. 1),
the size of founding populations also depends on birth
rate. While our finding that variation in genetic differen-
tiation is more sensitive to dispersal distance than birth
rate could be attributed to the range of treatments used
for these two parameters, it should be noted that birth
rate affects founding population size linearly. Dispersal
distance, on the other hand, operates in two dimensions
and the inverse square law ensures that its effect is non-
linear. For birth rate to be as sensitive as dispersal dis-
tance would require either an approximate threefold
increase in birth rates or a halving of the maximum dis-
persal distance.
Implications of spatial and temporal
variability in genetic patterns
The spatial and temporal variability in HS and FST result-
ing from disturbance regimes has some important impli-
cations for the measurement of neutral genetic patterns
and the inference of biological processes. For instance,
impacts of dispersal distance on FST were greater under
scenarios of large and/or severe disturbance compared to
undisturbed populations (Fig. 4A,B). For a given dispersal
pattern, the relationships between pairwise FST and geo-
graphic distance were substantially different in the pres-
ence of strong disturbance compared to undisturbed
populations. These changes were not represented by sim-
ple landscape-wide Mantel correlation statistics (which
test for the linear association between genetic and geo-
graphic distance), but by: (1) changes to overall FST,
Mantel r correlations within distance classes (which
enables visualization of spatial variation in the association
between genetic and geographic distance) (Fig. 4D), and
(2) distograms, which allow exploration of changes in the
magnitude of pairwise FST by distance class (Fig. 4C).
1192
I. D. Davies et al.
The expected linear relationship between ln-distance and
FST/(1–FST) was apparent for undisturbed populations.
However, even in IBD models of undisturbed popula-
tions, genetic drift will counter spatial patterns generated
by gene flow at a certain distance threshold (van Strien
et al. 2015). This is shown by our model (Fig. 4A), where
at a distance of 35 cells, the transformed FST no longer
responded to ln-distance in the same manner. For scenar-
ios in our model representing highly disturbed condi-
tions, the linear IBD relationship is disrupted over much
shorter geographic distances, indicating a stronger role of
drift in driving spatial genetic patterns in disturbance-
prone ecosystems. Similar effects on IBD patterns (after
single fire events) have been observed empirically in
response to fire history in several species of Florida sand
skink populations (Schrey et al. 2011). This presents
problems for the inference of dispersal processes from
spatial genetic data where disturbance history is
unknown. However, the highly variable IBD results
observed in Fig. 4B (MD = 0.25), occur only for scenar-
ios which result in relatively low population numbers
producing a spatially dynamic pattern of isolated popula-
tions (e.g., Fig. 6).
The temporal variability in H S and FST driven by dis-
turbance regimes (primarily disturbance size and severity
[Fig. 2C]), highlights how interpretations of “snapshot”
samples of genetic diversity need to consider this tempo-
ral variability when disturbance history is unknown.
Under some conditions (e.g., extreme population isola-
tion), the effects of single disturbance events can be
observed in populations over millennial timescales
(Beheregaray et al. 2003). Sampling designs conducted
without knowledge of historical disturbance regimes may
provide limited or misleading insights into contemporary
processes (Dyer et al. 2010; Landguth et al. 2010). A valu-
able area of further research may involve the use of simu-
lation tools to provide insights into the ability of
empirical sampling designs to recover the known patterns
of genetic diversity and dynamics from a simulation
model. Simulation tools (possibly in an Approximate
Bayesian Computation framework) may help with this.
Indeed, such model-based inference is increasingly com-
mon in molecular ecology, so considering disturbance
scenarios in the ecological-demographic-genetic models
used for these purposes is a logical approach to dealing
with this problem.
Toward increasing complexity
There is considerable scope for greater complexity in rep-
resentation of key mechanisms in future studies, including
representation of ecological, disturbance and genetic pro-
cesses.
ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd.
I. D. Davies et al.
Ecological representation
Successional dynamics are a well-documented conse-
quence of disturbance (Bengtsson et al. 2000; Lake et al.
2007; Leavesley et al. 2010; Banks et al. 2011), with spe-
cies commonly showing preference for early or late post-
disturbance stages, or having habitat suitability mediated
by disturbance return intervals. Therefore, disturbance
history can have major effects on amount and connectiv-
ity of suitable habitat, as well as viability of populations
(Amarasekare and Possingham 2001). Introducing simula-
tion approaches that explicitly include habitat dynamics
would further improve insights into dynamics of genetic
diversity under variation in disturbance regimes.
In addition, the basic population processes of many
species may not be static in response to disturbance
regimes. For instance, the reproductive strategies of many
plants are tied to disturbance (Gill 1975), and animal dis-
persal behavior and population connectivity may be vari-
able in response to recent disturbance history (Templeton
et al. 2001; Berry et al. 2005; Pereoglou et al. 2013).
These responses will vary with the degree of phenotypic
plasticity exhibited by the species in question (Anderson
et al. 2012).
Disturbance modelling
Significant scope exists to model specific disturbance
types more mechanistically using simulation methods that
vary with respect to complexity, stochasticity and process
representation (Bates and De Roo 2000; Keane et al.
2004; Seidl et al. 2011a). For instance, a consideration of
dispersal distance of organisms suggests that disturbance
shape, among other factors such as fire patchiness and
minimum fire return interval, is also critically important
given characteristics such as length-to-breadth ratio of
disturbance events will influence the rapidity of recolo-
nization dynamics and hence dynamics in genetic diver-
sity.
Assumptions about genetic processes
There are two further areas where this study might be
extended. Firstly, gene flow patterns are strongly tied to
demographic processes in our model, but there are sce-
narios where gene flow can be decoupled from dispersal
and colonization processes where, for example, animal
dispersal is sex-biased or plants show major differences in
seed and pollen dispersal capabilities. Secondly, selection-
driven genetic responses to disturbance history have been
documented (Moreira et al. 2014). Recent theoretical
models have focussed on evolution under spatio-temporal
ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd.
Disturbance and Genetic Diversity
environmental heterogeneity (Blanquart and Gandon
2011), and empirical studies quantifying selective effects
of disturbance regimes on traits and genes would be valu-
able for extending the present study.
Conclusions
Our findings highlight the background conditions under
which species must evolve strategies to persist in an envi-
ronment of recurrent disturbance. Key to this are survival
and dispersal strategies within the context of the temporal
and spatial scale of patterns of disturbance regimes. Dis-
turbance size, intensity, and frequency present conditions
within which life-history parameters, dispersal strategies,
and density-dependent behaviors may serve to limit or
enhance genetic diversity within in situ residual popula-
tions and at the invasion front, as these are the basic
mechanisms driving patterns of genetic diversity in an
environment subject to recurrent disturbance.
Acknowledgments
The authors would like to thank Jacques Gignoux and
Shayne Flint for the development of the software frame-
work in which this model was developed. The software is
available as Appendix S3 and will be available on GitHub
(https://github.com/) in the near future. In addition we
wish to thank Sam Yeaman and three anonymous review-
ers for valuable and insightful suggestions on ways to
improve this work.
Conflict of Interest
None declared.
References
Amarasekare, P., and H. Possingham. 2001. Patch dynamics
and metapopulation theory: the case of successional species.
J. Theor. Biol. 209:333–344.
Anderson, J. T., D. W. Inouye, A. M. McKinney, R. I.
Colautti, and T. Mitchell-Olds. 2012. Phenotypic plasticity
and adaptive evolution contribute to advancing flowering
phenology in response to climate change. Proc. Biol. Sci.
279:3843–3852
Banks, S. C., M. Dujardin, L. McBurney, D. Blair, M. Barker,
and D. B. Lindenmayer. 2011. Starting points for small
mammal population recovery after wildfire: recolonisation
or residual populations? Oikos 120:26–37.
Banks, S. C., G. J. Cary, A. L. Smith, I. D. Davies, D. A.
Driscoll, A. M. Gill, et al. 2013. How does ecological
disturbance influence genetic diversity? Trends Ecol. Evol.
28:670–679.
1193
Disturbance and Genetic Diversity
Banks, S. C., T. Lorin, R. E. Shaw, L. McBurney, D. Blair, M.
D. Blyton, et al. 2015. Fine-scale refuges can buffer
demographic and genetic processes against short-term
climatic variation and disturbance: a 22-year case study of
an arboreal marsupial. Mol. Ecol. 24:3831–3845.
Bates, P., and A. De Roo. 2000. A simple raster-based model
for flood inundation simulation. J. Hydrol. 236:54–77.
Beheregaray, L. B., C. Ciofi, D. Geist, J. P. Gibbs, A. Caccone,
and J. R. Powell. 2003. Genes record a prehistoric volcano
eruption in the Gal
apagos. Science 302:75–75.
Bengtsson, J., S. G. Nilsson, A. Franc, and P. Menozzi. 2000.
Biodiversity, disturbances, ecosystem function and
management of European forests. For. Ecol. Manage.
132:39–50.
Berry, O., M. D. Tocher, D. M. Gleeson, and S. D. Sarre. 2005.
Effect of vegetation matrix on animal dispersal: genetic
evidence from a study of endangered skinks. Conserv. Biol.
19:855–864.
Berumen, M. L., and M. S. Pratchett. 2006. Recovery without
resilience: persistent disturbance and long-term shifts in the
structure of fish and coral communities at Tiahura Reef,
Moorea. Coral Reefs 25:647–653.
Blanquart, F., and S. Gandon. 2011. Evolution of migration in
a periodically changing environment. Am. Nat. 177:188–201.
Boer, M. M., R. J. Sadler, R. A. Bradstock, A. M. Gill, and P.
F. Grierson. 2008. Spatial scale invariance of southern
Australian forest fires mirrors the scaling behaviour of fire-
driving weather events. Landsc. Ecol. 23:899–913.
Boer, M. M., R. J. Sadler, R. S. Wittkuhn, L. McCaw, and P. F.
Grierson. 2009. Long-term impacts of prescribed burning on
regional extent and incidence of wildfires – evidence from
50 years of active fire management in SW Australian forests.
For. Ecol. Manage. 259:132–142.
Bradstock, R., G. Cary, I. Davies, D. Lindenmayer, O. Price,
and R. Williams. 2012. Wildfires, fuel treatment and risk
mitigation in Australian eucalypt forests: insights from
landscape-scale simulation. J. Environ. Manage. 105:66–75.
Byram, G. M. 1959. Combustion of forest fuels. Pp. 61–89 in
K. P. Davis, ed. Forest fire: control and use, vol. 1, 61–89.
Cary, G. J., R. E. Keane, R. H. Gardner, S. Lavorel, M. D.
Flannigan, I. D. Davies, et al. 2006. Comparison of the
sensitivity of landscape-fire-succession models to variation
in terrain, fuel pattern, climate and weather. Landsc. Ecol.
21:121–137.
Cary, G., R. A. Bradstock, A. M. Gill, and R. J. Williams. 2012.
Global change and fire regimes in Australia. Pp. 149–169 in
R. A. Bradstock, R. J. Williams, A. M. Gill, ed. Flammable
Australia: fire regimes biodiversity and ecosystems in a
changing world. CSIRO Publishing, Australia.
Connell, J. H.. 1980. Diversity and the coevolution of
competitors, or the ghost of competition past. Oikos
35:131–138.
Connell, J. H., and R. O. Slatyer. 1977. Mechanisms of
succession in natural communities and their role in
1194
I. D. Davies et al.
community stability and organization. Am. Nat. 111:1119–
1144.
Cumming, S. 2001. A parametric model of the fire-size
distribution. Can. J. For. Res. 31:1297–1303.
Diniz-Filho, J. A. F., T. N. Soares, J. S. Lima, R. Dobrovolski,
V. L. Landeiro, M. P. d. C. Telles, et al. 2013. Mantel test in
population genetics. Genet. Mol. Biol. 36:475–485.
Dyer, R. J., J. D. Nason, and R. C. Garrick. 2010. Landscape
modelling of gene flow: improved power using conditional
genetic distance derived from the topology of population
networks. Mol. Ecol. 19:3746–3759.
Excoffier, L., and N. Ray. 2008. Surfing during population
expansions promotes genetic revolutions and structuration.
Trends Ecol. Evol. 23:347–351.
Feller, M., 1996. The influence of fire severity, not fire intensity,
on understory vegetation biomass in British Columbia, 13th
Conference on Fire and Meteorology, Lorne, Australia.
Flannigan, M. D., M. A. Krawchuk, W. J. de Groot, B. M.
Wotton, and L. M. Gowman. 2009. Implications of
changing climate for global wildland fire. Int. J. Wildl. Fire
18:483–507.
Foster, D. R., D. H. Knight, and J. F. Franklin. 1998.
Landscape patterns and legacies resulting from large,
infrequent forest disturbances. Ecosystems 1:497–510.
Franklin, J. F., and R. T. Forman. 1987. Creating landscape
patterns by forest cutting: ecological consequences and
principles. Landsc. Ecol. 1:5–18.
Gardner, R. H., B. T. Milne, M. G. Turnei, and R. V. O’Neill.
1987. Neutral models for the analysis of broad-scale
landscape pattern. Landsc. Ecol. 1:19–28.
Gibbs, J. P. 2001. Demography versus habitat fragmentation as
determinants of genetic variation in wild populations. Biol.
Conserv. 100:15–20.
Gill, A. M. 1975. Fire and the Australian flora: a review. Aust.
For. 38:4–25.
Gill, A. M., S. L. Stephens, and G. J. Cary. 2013. The
worldwide “wildfire” problem. Ecol. Appl. 23:438–454.
Grimm, V., U. Berger, D. L. DeAngelis, J. G. Polhill, J. Giske,
and S. F. Railsback. 2010. The ODD protocol: a review and
first update. Ecol. Model. 221:2760–2768.
Haefner, J. W., G. C. Poole, P. V. Dunn, and R. T. Decker.
1991. Edge effects in computer models of spatial
competition. Ecol. Model. 56:221–244.
Hallatschek, O., P. Hersen, S. Ramanathan, and D. R. Nelson.
2007. Genetic drift at expanding frontiers promotes
gene segregation, Proc. Natl Acad. Sci. USA 104:19926–
19930.
Helmus, M. R., W. B. Keller, M. J. Paterson, N. D. Yan, C. H.
Cannon, and J. A. Rusak. 2010. Communities contain
closely related species during ecosystem disturbance. Ecol.
Lett. 13:162–174.
Hughes, T., and J. Connell. 1999. Multiple stressors on coral
reefs: a long-term perspective. Limnol. Oceanogr. 44
(3part2):932–940.
ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd.
I. D. Davies et al.
Jarne, P., and P. J. Lagoda. 1996. Microsatellites, from
molecules to populations and back. Trends Ecol. Evol.
11:424–429.
Keane, R. E., G. J. Cary, I. D. Davies, M. D. Flannigan, R. H.
Gardner, S. Lavorel, et al. 2004. A classification of landscape
fire succession models: spatial simulations of fire and
vegetation dynamics. Ecol. Model. 179:3–27.
Keeley, J. E. 2009. Fire intensity, fire severity and burn
severity: a brief review and suggested usage. Int. J. Wildl.
Fire 18:116–126.
King, K. J., G. J. Cary, R. A. Bradstock, and J. B. Marsden-
Smedley. 2013. Contrasting fire responses to climate and
management: insights from two Australian ecosystems.
Glob. Change Biol. 19:1223–1235.
Koerner, S. E., and S. L. Collins. 2014. Interactive effects of
grazing, drought, and fire on grassland plant communities
in North America and South Africa. Ecology 95:98–109.
Krebs, P., G. B. Pezzatti, S. Mazzoleni, L. M. Talbot, and M.
Conedera. 2010. Fire regime: history and definition of a key
concept in disturbance ecology. Theory Biosci. 129:53–69.
Lake, P. 2000. Disturbance, patchiness, and diversity in
streams. J. N. Am. Benthol. Soc. 19:573–592.
Lake, P., N. Bond, and P. Reich. 2007. Linking ecological
theory with stream restoration. Freshw. Biol. 52:597–615.
Landguth, E., S. Cushman, M. Schwartz, K. McKelvey, M.
Murphy, and G. Luikart. 2010. Quantifying the lag time to
detect barriers in landscape genetics. Mol. Ecol. 19:4179–
4191.
Leavesley, A. J., G. J. Cary, G. P. Edwards, and A. M. Gill.
2010. The effect of fire on birds of mulga woodland in arid
central Australia. Int. J. Wildl. Fire 19:949–960.
Lindenmayer, D. B., R. J. Hobbs, G. E. Likens, C. J. Krebs, and
S. C. Banks. 2011. Newly discovered landscape traps
produce regime shifts in wet forests. Proc. Natl Acad. Sci.
USA 108:15887–15891.
Lirman, D. 2003. A simulation model of the population
dynamics of the branching coral Acropora palmata Effects
of storm intensity and frequency. Ecol. Modell. 161:169–
182.
Lowe, W. H., and F. W. Allendorf. 2010. What can genetics
tell us about population connectivity? Mol. Ecol. 19:3038–
3051.
Mantel, N., 1967. The detection of disease clustering and a
generalized regression approach. Cancer Res. 27(2 Part 1):
209–220.
Miller, A. D., S. H. Roxburgh, and K. Shea. 2011. How
frequency and intensity shape diversity–disturbance
relationships. Proc. Natl Acad. Sci. USA 108:5643–5648.
Moreira, B., M. C. Castellanos, and J. Pausas. 2014. Genetic
component of flammability variation in a Mediterranean
shrub. Mol. Ecol. 23:1213–1223.
Nei, M. 1978. Estimation of average heterozygosity and genetic
distance from a small number of individuals. Genetics
89:583–590.
ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd.
Disturbance and Genetic Diversity
Noble, I. R., and R. Slatyer, 1980. The use of vital attributes to
predict successional changes in plant communities subject to
recurrent disturbances. Vegetatio 43:5–21.
Pannell, J. R., and B. Charlesworth. 2000. Effects of
metapopulation processes on measures of genetic diversity.
Philos. Trans. R. Soc. Lond. B Biol. Sci. 355:1851–1864.
Pereoglou, F., D. Lindenmayer, C. MacGregor, F. Ford, J.
Wood, and S. Banks. 2013. Landscape genetics of an early
successional specialist in a disturbance-prone environment.
Mol. Ecol. 22:1267–1281.
Petit, R. J., I. Aguinagalde, J.-L. de Beaulieu, C. Bittkau, S.
Brewer, R. Cheddadi, et al. 2003. Glacial refugia: hotspots but
not melting pots of genetic diversity. Science 300:1563–1565.
R Core Team, 2014. R: A language and environment for
statistical computing. R Foundation for Statistical
Computing, Vienna, Austria. ISBN 3-900051-07-0.
Raffa, K. F., B. H. Aukema, B. J. Bentz, A. L. Carroll, J. A. Hicke,
M. G. Turner, et al. 2008. Cross-scale drivers of natural
disturbances prone to anthropogenic amplification: the
dynamics of bark beetle eruptions. Bioscience 58:501–517.
Romiguier, J., P. Gayral, M. Ballenghien, A. Bernard, V.
Cahais, A. Chenuil, et al. 2014. Comparative population
genomics in animals uncovers the determinants of genetic
diversity. Nature 515:261–263.
Romme, W. H., E. H. Everham, L. E. Frelich, M. A. Moritz,
and R. E. Sparks. 1998. Are large, infrequent disturbances
qualitatively different from small, frequent disturbances?
Ecosystems 1:524–534.
Rousset, F. 1997. Genetic differentiation and estimation of
gene flow from F-statistics under isolation by distance.
Genetics 145:1219–1228.
Schrey, A. W., K. G. Ashton, S. Heath, E. D. McCoy, and H.
R. Mushinsky. 2011. Fire alters patterns of genetic diversity
among 3 lizard species in Florida scrub habitat. J. Hered.
102:399–408.
Seidl, R., P. M. Fernandes, T. F. Fonseca, F. Gillet, A. M.
J€
onsson, K. Merganicov
a, et al. 2011a. Modelling natural
disturbances in forest ecosystems: a review. Ecol. Model.
222:903–924.
Seidl, R., M. J. Schelhaas, and M. J. Lexer. 2011b. Unraveling
the drivers of intensifying forest disturbance regimes in
Europe. Glob. Chang. Biol. 17:2842–2852.
Shea, K., S. H. Roxburgh, and E. S. Rauschert. 2004. Moving
from pattern to process: coexistence mechanisms under
intermediate disturbance regimes. Ecol. Lett. 7:491–508.
Slatkin, M. 1977. Gene flow and genetic drift in a species
subject to frequent local extinctions. Theor. Popul. Biol.
12:253–262.
Sousa, W. P. 1984. The role of disturbance in natural
communities. Annu. Rev. Ecol. Syst. 15:353–391.
Spear, S. F., and A. Storfer. 2010. Anthropogenic and natural
disturbance lead to differing patterns of gene flow in the
Rocky Mountain tailed frog Ascaphus montanus. Biol.
Conserv. 143:778–786.
1195
Disturbance and Genetic Diversity
Spies, T. A., D. B. Lindenmayer, A. M. Gill, S. L. Stephens,
and J. K. Agee. 2012. Challenges and a checklist for
biodiversity conservation in fire-prone forests: Perspectives
from the Pacific Northwest of USA and Southeastern
Australia. Biol. Conserv. 145:5–14.
van Strien, M. J., R. Holderegger, and H. J. Van Heck. 2015.
Isolation-by-distance in landscapes: considerations for
landscape genetics. Heredity 114:27–37.
Su
arez, N., E. Betancor, R. Fregel, F. Rodriguez, and J.
Pestano. 2012. Genetic signature of a severe forest fire on
the endangered Gran Canaria blue chaffinch (Fringilla teydea
polatzeki). Conserv. Genet. 13:499–507.
Swetnam, T. W., and J. L. Betancourt, 2010. Mesoscale
disturbance and ecological response to decadal climatic
variability in the American Southwest. Pp. 329–359 in
M. Stoffel, M. Bollschweiler, D.R. Butler and B.H. Luckman
ed. Tree rings and natural hazards. Springer, Netherlands.
Templeton, A. R., R. J. Robertson, J. Brisson, and J. Strasburg.
2001. Disrupting evolutionary processes: the effect of habitat
fragmentation on collared lizards in the Missouri Ozarks.
Proc. Natl Acad. Sci. USA 98:5426–5432.
Thodsen, H., A. Baattrup-Pedersen, H. E. Andersen, K. M. B.
Jensen, P. M. Andersen, K. Bolding, et al., 2014. Climate
change effects on lowland stream flood regimes and riparian
rich fen vegetation communities in Denmark. Hydrol. Sci. J.
URL: http://dx.doi.org/10.1080/02626667.2014.990965.
Turner, M. G. 2010. Disturbance and landscape dynamics in a
changing world 1. Ecology 91:2833–2849.
Turner, M. G., W. H. Romme, R. H. Gardner, R. V. O’Neill,
and T. K. Kratz. 1993. A revised concept of landscape
equilibrium: disturbance and stability on scaled landscapes.
Landsc. Ecol. 8:213–227.
Wade, M. J., and D. E. McCauley. 1988. Extinction and
recolonization: their effects on the genetic differentiation of
local populations. Evolution 42:995–1005.
Waters, J. M., C. I. Fraser, and G. M. Hewitt. 2013. Founder
takes all: density-dependent processes structure biodiversity.
Trends Ecol. Evol. 28:78–85.
1196
I. D. Davies et al.
Webster, P. J., G. J. Holland, J. A. Curry, and H.-R. Chang.
2005. Changes in tropical cyclone number, duration, and
intensity in a warming environment. Science 309:1844–1846.
Whelan, R., L. Rodgerson, C. R. Dickman, and E. F.
Sutherland. 2002. Critical life cycles of plants and animals:
developing a process-based understanding of population
changes in fire-prone landscapes. Pp. 94–124 in R. A.
Bradstock, R. J. Williams and A. M. Gill ed. Flammable
Australia: the fire regimes and biodiversity of a continent.
Cambridge University Press, Cambridge.
White, J. W., A. Rassweiler, J. F. Samhouri, A. C. Stier, and C.
White. 2014. Ecologists should not use statistical significance
tests to interpret simulation model results. Oikos 123:385–
388.
Whitlock, M. C., and D. E. McCauley. 1990. Some population
genetic consequences of colony formation and extinction:
genetic correlations within founding groups. Evolution
44:1717–1724.
Supporting Information
Additional Supporting Information may be found in the
online version of this article:
Appendix S1. Additional quantification of the relative
importance of seven other experimental treatments
referred to in this article: landscape size; landscape topol-
ogy; variable disturbance sizes; carrying capacity; negative
exponential dispersal kernel; the number of alleles studied
and the time during the annual cycle, when the popula-
tion is limited to carrying capacity.
Appendix S2. Overview, Details and Design concepts
(ODD) of the model used in this article following the
suggested protocol by (Grimm et al. 2010).
Appendix S3. Java source code, parameter and input files
together with installation instructions for the model used
in article.
ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd.